Efficiency of fall and spring broadcast fertilizer phosphorus application for corn and soybean in no-till

Transcription

1 Retrospective Theses and Dissertations 2007 Efficiency of fall and spring broadcast fertilizer phosphorus application for corn and soybean in no-till Sebastián Raúl Barcos Iowa State University Follow this and additional works at: Part of the Agricultural Science Commons, Agriculture Commons, Agronomy and Crop Sciences Commons, and the Soil Science Commons Recommended Citation Barcos, Sebastián Raúl, "Efficiency of fall and spring broadcast fertilizer phosphorus application for corn and soybean in no-till" (2007). Retrospective Theses and Dissertations This Thesis is brought to you for free and open access by Iowa State University Digital Repository. It has been accepted for inclusion in Retrospective Theses and Dissertations by an authorized administrator of Iowa State University Digital Repository. For more information, please contact

2 Efficiency of fall and spring broadcast fertilizer phosphorus application for corn and soybean in no-till by Sebastián Raúl Barcos A thesis submitted to the graduate faculty in partial fulfillment of the requirements for the degree of MASTER OF SCIENCE Major: Soil Science (Soil Fertility) Program of Study Committee: Antonio P. Mallarino, Major Professor Russ Mullen John Sawyer Iowa State University Ames, Iowa 2007 Copyright Sebastián Raúl Barcos, All rights reserved.

3 UMI Number: UMI Microform Copyright 2008 by ProQuest Information and Learning Company. All rights reserved. This microform edition is protected against unauthorized copying under Title 17, United States Code. ProQuest Information and Learning Company 300 North Zeeb Road P.O. Box 1346 Ann Arbor, MI

4 ii To: my father and my mother

5 iii TABLE OF CONTENTS LIST OF TABLES iv LIST OF FIGURES v CHAPTER 1: GENERAL INTRODUCTION 1 THESIS ORGANIZATION 3 CHAPTER 2: EFFICIENCY OF FALL AND SPRING BROADCAST 4 FERTILIZER PHOSPHORUS APPLICATION FOR CORN AND SOYBEAN IN NO-TILL ABSTRACT 4 INTRODUCTION 5 MATERIALS AND METHODS 9 Sites and treatments 9 Statistical analysis and evaluation of crop response to the treatments 13 RESULTS AND DISCUSSIONS 14 Corn and soybean grain yield responses 14 Corn and soybean early dry weight responses 17 Corn and soybean early P concentration responses 20 Corn and soybean early P uptake responses 22 Summary discussion 24 CONCLUSIONS 26 REFERENCES 27 CHAPTER 3: GENERAL CONCLUSIONS 45 ACKNOWLEDGEMENTS 47

6 iv LIST OF TABLES Table 1. Location, soils, tillage history, planting date, and reside cover. 31 Table 2. Soil-test P (Bray-1, Mehlich-3, Olsen), ph, and organic matter for all sites. 32 Table 3. Grain yield as affected by the P fertilization rates and time of broadcast P 33 application. Table 4. Models fit to relationships between grain yield and P rates for responsive 34 sites, and estimated P rate for maximum yield response. Table 5. Early plant growth (V5 to V6 stage) as affected by the P fertilization rate 35 and the time of broadcast P application. Table 6. Models fit to relationships between early plant weight and P rates for 36 responsive sites, and estimated P rate for maximum yield response. Table 7. Early plant P concentration (V5 to V6 stage) as affected by the P 37 fertilization rate and the time of broadcast P application. Table 8. Models fit to relationships between early plant P concentration and P rates 38 for responsive sites, and estimated P rate for maximum yield response. Table 9. Early plant P uptake (V5 to V6 stage) as affected by the P fertilization rate 39 and the time of broadcast P application. Table 10. Models fit to relationships between early plant P uptake and P rates for 40 responsive sites, and estimated P rate for maximum yield response.

7 v LIST OF FIGURES Fig. 1. Mean grain yield response to P fertilization across responsive sites. 41 Fig. 2. Mean early plant growth response to P fertilization across responsive sites. 42 Fig. 3. Mean early plant P concentration response to P fertilization across 43 responsive sites. Fig. 3. Mean early plant P uptake response to P fertilization across responsive sites. 44

8 1 CHAPTER 1: GENERAL INTRODUCTION Conservational tillage systems (especially no-till) offer several potential advantages over conventional tillage methods. Usually soil moisture is increased and water loss is reduced, soil loss with various erosion mechanisms is reduced, and root activity can increase in drier periods resulting in improved nutrient uptake efficiency of shallow roots during summer. Phosphorus is an essential nutrient for crops and has limited movement in soil because it is more tightly retained than other nutrients. Therefore, conservation tillage may also affect P distribution and plant availability of P in soils and the efficiency of P fertilizer application methods. Soil P testing and estimates of P removal with harvest are two key tools to asses the correct management for this nutrient. Survey data suggest that approximately 70 % of the soils in Iowa test Optimum (16 to 20 mg P kg -1 ) to Very High in P. Moreover, soil P levels may be increasing because many farmers do not want to risk P deficiency. The usual P management in Iowa for a corn-soybean rotation under tillage or no-till systems is to broadcast the P fertilizer in the fall, most of the time applying the P fertilizer for the two crops at once, always before the corn. Lower fertilizer application costs than for band placements, drier soils than in early spring that minimizes soil compaction by farm equipment, and time to do it before snowfall and freezing of the soils, are some of the reasons P is manage this way. Very fine texture, some clay types, and very acid or highly calcareous soils, can lead soils to tightly retain P when fertilizers are mixed with soil. In these conditions banding may

9 2 improve P efficiency, P uptake, and yield. But soil properties and placement research in Iowa suggest banding seldom improves yield. However, no-till management and broadcast P fertilization usually lead to P stratification in soils. Phosphorus is one of the more immobile nutrients (together with K, for example) in soils and is highly retained to soil constituents. Therefore, P does not substantially move vertically in a soil when it is applied. This soil P stratification could cause problems if the surface soil is dried, and consequently could reduce the capacity of plant roots to absorb P from the soil. Broadcasting P fertilizer in no-till systems can limit the uptake if the soil surface is dry and residue cover is too thick or too little, and can increase the risk of loss with runoff if applied to frozen or saturated soils. However, research in the Corn Belt and in other regions has shown that no-till management may reduce P sorption by soil constituents of surface soil layers. Moreover, research in Iowa on P fertilizer placement during the last decade at many fields showed small and inconsistent differences between banding and broadcast P application methods for corn and soybean. The small and inconsistent differences between the P placement methods in Iowa and often significant differences in other regions could be explained by differences in soil properties and climate, mainly P sorption capacity of the soils and both rainfall amounts and patterns. However, a lack of response to P placement methods for corn and soybean in Iowa and highly variable responses in other regions might also be explained in part by differences in the timing of the broadcast P application. In most Iowa experiments the P fertilizer was broadcast in the fall, while in experiments in other regions the P was broadcast in fall or spring. Weather, soil conditions, and the length of time

10 3 between fertilizer application and the planting of the crop could affect plant-availability of P, and P uptake by roots, and grain yield. Therefore, the goal of this thesis research was to confirm or disprove that fall broadcast P fertilization for corn and soybean managed in no-tillage is more efficient than spring broadcast P fertilization. The efficiency of this study will be based on measurements of early plant growth, early P uptake, and grain yield responses. THESIS ORGANIZATION This thesis is presented as one paper suitable for publication in scientific journals of the American Society of Agronomy or Soil Science Society of America. The paper is entitled Efficiency of Fall and Spring Broadcast Fertilizer Phosphorus Application for Corn and Soybean in No-Till. The paper includes sections for an abstract, introduction, materials and methods, results and discussion, conclusions, references, tables, and figures. The paper is preceded by a general introduction and is followed by a general conclusions section.

11 4 CHAPTER 2: EFFICIENCY OF FALL AND SPRING BROADCAST FERTILIZER PHOSPHORUS APPLICATION FOR CORN AND SOYBEAN IN NO-TILL A paper to be submitted to a Journal of the America Society of Agronomy Sebastián R. Barcos and Antonio P. Mallarino ABSTRACT Previous Iowa research during many years has shown no large or consistent grain yield differences among broadcast, shallow band, and deep band P placement methods for no-till corn (Zea mays L.) and soybean (Glycine max [L.] Merr.). Lower efficiency for broadcast P sometimes has been shown in other regions. One reason for lack of differences between P placement methods in Iowa might be that broadcast P was always applied in the fall (4-5 months before planting) with sufficient time to reach soil and shallow roots by the time it was needed by plants. This study evaluated the efficiency of fall and spring broadcast P application for corn and soybean under no-till by conducting 20 trials during 2005, 2006, and 2007 on fields testing 6 to 29 mg kg -1 (Bray-P1, 15-cm depth). Triple superphosphate was broadcast at 0, 10, 20, 30, 40, and 50 kg P ha -1 in the fall (in November or early December) and in spring (7 to 10 d prior to planting). We measured initial soil P (0-7.5 and cm) with three soil-test methods (Bray-P1, Olsen, and Mehlich-3), early plant growth, early P

12 5 concentration, and P uptake (at the V5-V6 stage), and grain yield. Significant crop responses to P were observed at 11 sites for grain yield, three sites for early growth, six sites for early P concentration, and seven sites for early P uptake. The time of P application did not affect grain yield at any site. There were small and inconsistent or unreasonable time of application effects at one site for early growth, one site for early P concentration, and one site for early P uptake. Perhaps Iowa soil properties and usually humid climate explain the lack of difference among P placement methods for these crops. INTRODUCTION Phosphorus is an essential nutrient for crops, and soil P testing is a useful diagnostic tool to determine its sufficiency level in soils. Several studies in Iowa and the northern Corn Belt showed that in predominant soils the probability of grain yield increases from P fertilization of corn or soybean is large only in low-testing soils (< 21 mg P kg -1 by the Bray-P1 extractant) (demooy et al., 1973; Bharati et al., 1986; Rehm, 1986; Mallarino et al., 1991; Mallarino and Blackmer, 1992; Webb et al., 1992; Mallarino, 1997; Randall et al., 1997; Mallarino, 2003, Dodd and Mallarino, 2005). Iowa fertilization guidelines recommend only maintenance fertilization for soils testing optimum (16 to 20 mg P kg -1 ) and no fertilization other than small starter fertilizer rates for some specific conditions in high-testing soils (Sawyer et al., 2002). Over the years there has been a steady increase in soil-test P (STP) levels in Iowa, however, and many fields are being fertilized in spite of high soil-test levels

13 6 because many farmers do not want to risk P deficiency. Application of P fertilizer for crops in high-testing soils reduces cropping profitability and may also have negative environmental consequences. An increase in the amount of P in surface waters because of excess P loss from agricultural fields has received public attention in recent years. The usual P management for the corn-soybean rotation in Iowa is to broadcast the P fertilizer in the fall for both tillage and no-tillage systems. Several practical reasons encourage this management practice; such as lower fertilizer application costs than for band placements, drier soils than in early spring that minimize soil compaction by farm equipment, and time to do it before snowfall and freeze of the soils. No-till management and broadcast P fertilization usually lead to P stratification in soils. Phosphorus accumulates at or near the soil surface as a result of a minimal mixing of crop residues and surface-applied fertilizers with soil, restricted movement of P through soil layers, and a slow cycling of nutrients from deep soil layers to shallow layers through nutrient uptake by roots (Shear and Moschler, 1969; Griffith et al., 1977; Mackay et al., 1987; Karlen et al., 1991). In dry periods, however, a relative accumulation of P near the soil surface may decrease nutrient availability to plants. Usually soil moisture is increased and soil temperature is reduced with a high residue coverage in no-till soils at shallow depths, which can inhibit plant growth and nutrient availability early in the season but can increase root activity in drier periods (Barber, 1971; Al-Darby and Lowery, 1978; Fortin, 1993). No-till management and broadcast P fertilization can also affect soil P sorption and, therefore, plant P availability. Phosphorus sorption by soil constituents is reduced in surface layers of no-till soils (Guertal et al., 1991). Several cations, but mainly Al, Ca, and Fe, are

14 7 highly reactive with applied water-soluble P and can reduce the plant availability of the P if soil conditions are appropriate. For example, the plant availability of water-soluble P forms may be quickly and significantly reduced by soil CaCO 3 in high-ph soils and by Al and Fe if soil ph is strongly acidic. The research by Guertal et al. (1991) and others suggested that higher levels or organic matter in the shallow layers of no-till soils reduces some of these effects and may benefit crop P uptake. Placing the P fertilizer in subsurface bands reduces P accumulation in the near-surface layers of soil and may affect P availability and uptake by crops. Singh et al. (1966) and Molscher and Martens (1975) in Virginia, and Belcher and Ragland (1972) in Kentucky, showed that absorption of P by no-till corn when the fertilizer was banded onto the soil surface was equal to or better than absorption when P was broadcast and incorporated. The successful use of the surface bands was mainly attributed to a greater root activity in the shallower soil layers due to higher soil moisture content under the residue and the adequate rainfall received during the growing season. Hairston et al. (1990), in Mississippi, showed that deep injection (15-cm depth) of P fertilizer compared to broadcast applications had a superior yield response on no-till soybean in soils testing low in P, while conventionally tilled soybean responded similarly to all placement methods. Researchers in the humid areas of the Corn Belt have found inconsistent decreases in P availability for corn or soybean due to soil P stratification and broadcast P fertilization compared to other placement methods. Eckert and Johnson (1985) compared several rates of P fertilizer broadcast on the surface with subsurface banding (5 cm on the side and below the seed) for no-till corn in Ohio soils and concluded that the subsurface banding resulted in

15 8 greater fertilizer use efficiency. They also concluded that higher broadcast fertilizer rates and higher STP levels were needed to achieve P sufficiency in no-till management compared with conventional tillage. This result coincides with known effects of banding in minimizing retention of this nutrient by soil constituents and sometimes increasing fertilizer use efficiency by crops (Black, 1993). However, Mengel et al. (1988), in Indiana, reported that the placement of pre-plant P fertilizer (broadcast, stripped onto the soil surface, or deepbanded 20 cm deep) had no significant effect on corn yield in either tillage system (plow and no-till). Iowa research with no-till corn (Bordoli and Mallarino, 1998; Mallarino et al., 1999) showed that P banding increased early growth and P uptake but did not increase yield. Research in Iowa with no-till soybean by Borges and Mallarino (2000), Buah et al. (2000), and Borges and Mallarino (2001) showed that P fertilization often increased yield in lowtesting soils but band or broadcast placement methods did not differ. This Iowa research showed larger corn yield response to deep-band K fertilizer (15-20 cm below the ridge surface) compared with broadcast or planter-band application but little or no difference between P placement methods. The lack of differences between the fertilizer placement methods with no-till was also found for the chisel-plow tillage system. Such inconsistent results from research on P placement methods suggest that the response to broadcast or band P fertilizer should not be extrapolated over a wide range of soil and weather conditions. Furthermore, the timing of the broadcast P application varied across studies and might in part explain inconsistent differences between placement methods. With very few exceptions the broadcast P fertilizer applied in the reported Iowa studies was applied in the fall from 4 to 5 months before planting. This timing of broadcast application

16 9 was in contrast to spring application used in some studies (Moschler et al., 1971; Belcher and Ragland, 1972; Bharati et al., 1986; Eckert and Johnson, 1985; and Rehm, 1986) but similar to that used by Mackay et al. (1987) and by Rehm et al. (1995) for some treatments. The movement of P, both as fertilizer granules or dissolved by rainfall or snowfall, through the residue cover and the shallowest few cm of soil could be affected by the length of time, weather, and soil conditions during the period between the broadcast P application and planting of crops. These processes could affect P sorption by soil constituents, plantavailability of P, and P uptake by roots from the most shallow soil layers. The objective of this study was to evaluate the efficacy of fall and spring broadcast P fertilization for corn and soybean managed with no-tillage by measuring early plant growth, early P uptake, and grain yield responses to P. This research should provide information useful to assess if the common practice of broadcasting P application in the fall is effective and if a lack of large and consistent responses to P placement methods in Iowa was the result of broadcast P application in the fall. MATERIALS AND METHODS Sites and Treatments Twenty one-year trials with corn and soybean managed with no-tillage were established in Iowa. The experiments were conducted during 2005, 2006, and 2007 to evaluate grain yield, early plant growth, and early P uptake responses to broadcast P fertilization in fall or

17 10 spring. Sites with corn-soybean rotation histories were selected to represent a wide range of soil series, years of no-till management, and Very Low to High STP in the 15-cm surface layer according to Iowa State University interpretations (< 31 mg kg -1, Bray-P1). Table 1 shows summarized site information. Thirteen sites were established at four Iowa State University research farms located in Northwest Iowa (near Sutherland), in Southwest Iowa (near Atlantic), in Southeast Iowa (near Crawfordsville), and in central Iowa (near Boone). Seven trials were established at farmer s fields in several regions of the state. Corn and soybean were planted with equipment owned by the research farms or the farmers. Management practices were those normally used at each farm and, thus, corn hybrids, soybean varieties, seeding rates, row width, planting dates, herbicide management, and planting equipment varied among fields. In sites where the row spacing was 76 cm, the planters had residue managers that swept residue away from the row. The treatments (11) were a control receiving no P and the factorial combination of five P rates and two application times. Granulated triple superphosphate was broadcast by hand at rates of 10, 20, 30, 40, or 50 kg P ha -1. The times of P application were fall and spring. In the fall, treatments were applied in November to early December after crop harvest and before heavy snowfall or before soils froze. In spring, the P was applied 7 to 10 d prior to planting the crops. Randomized complete block designs with three replications were used for all trials. To better evaluate crop measurements for the control treatment, and to complete a rectangular grid of plots at each trial, two control plots were included in each block. Plot length ranged from 12 to 15 m across sites, and plot width ranged from 4.5 to 6 m depending on the planter width.

18 11 Table 2 shows summarized information about initial soil-test values of the sites. To measure initial STP and soil P stratification, a composite sample of 10 to 12 soil cores was taken from two depths (0-7.5 cm and cm) from the control plots prior to the fall treatment application. Samples were tested for P by the Bray-P1, Olsen, and Mehlich-3 tests using a colorimetric P determination method (Murphy and Riley, 1962), and for ph by using a 1:1 soil:water ratio. The laboratory procedures followed procedures recommended for the North Central Region by the North Central Region Soil and Plant Analysis Committee, NCR- 13 (Brown, 1998). Soil organic matter was measured using a combustion method (Wang and Anderson, 1998). Soil samples taken from the top 5-cm layer were analyzed for particle size analyses at the University of Nebraska Soil and Plant Analysis Laboratory by a method described by Kettler et al. (2001). The current STP interpretation for corn and soybean grain production is used in this paper to classify STP ranges (Sawyer et al., 2002). Boundary values for five STP classes for soil series classified as low in subsoil P for the Bray-P1 or Mehlich-3 tests with a colorimetric determination of extracted P are 8 mg kg -1 for Very Low, 9 to 15 mg kg -1 for Low, 16 to 20 mg kg -1 for Optimum, 21 to 30 mg kg -1 for High, and 31 mg kg -1 for Very High. When a series subsoil is classified as high in P content, these boundaries change to 5 mg kg -1 for Very Low, 6 to 10 mg kg -1 for Low, 11 to 15 mg kg -1 for Optimum, 16 to 20 mg kg -1 for High, and 21 mg kg -1 for Very High. Boundary values for soil series with low subsoil P for the Olsen P test are 5 mg kg -1 for Very Low, 6 to 10 mg kg -1 for Low, 11 to 14 mg kg -1 for Optimum, 15 to 20 mg kg -1 for High, and 21 mg kg -1 for Very High. When a series subsoil is classified as high in P content, these boundaries

19 12 change to 3 mg kg -1 for Very Low, 4 to 7 mg kg -1 for Low, 8 to 11 mg kg -1 for Optimum, 12 to 15 mg kg -1 for High, and 16 mg kg -1 for Very High. Corn and soybean early growth was determined by sampling the above-ground portions of 10 plants from crop rows that would not be harvested and that were not border rows at the V5 to V6 growth stage (Fehr et al., 1971; Ritchie et al., 1986). Plants were dried at 65 C in a forced-air oven, weighed, and ground to pass through a 2-mm sieve. Plant material was digested using a H 2 SO 4- H 2 O 2 method (Digesdahl Analysis System, Hach Inc., Boulder, CO) and P in extracts was measured by the Murphy and Riley (1962) colorimetric procedure. Early plant P uptake was calculated from plant P concentration and oven-dry weights. Corn and soybean were harvested with a plot combine at research farms and by hand at farmers' fields. To avoid plot border effects, three to four central rows were harvested at the research farms and two central rows were harvested at farmers' fields; and a 1.8- to 2.25-m border between plots along rows was not included in the harvest area. For the hand harvest, the corn ears and soybean plants cut at a ground level were threshed with a stationary plot combine. The grain was weighed, a subsample was collected for moisture determination, and corn and soybean yield were adjusted to 155 and 130 mg kg -1 H 2 O, respectively. Plant population was measured at harvest on two central rows of each plot (except at trials conducted in 2005), but data are not shown because the treatments did not affect (P 0.05) the plant population at any site.

20 13 Statistical Analysis and Evaluation of Crop Response to the Treatments Statistical analysis of grain yield, early plant growth, P concentration and P uptake were conducted for each site using the MIXED procedure of SAS (SAS Institute, 2000) for a randomized complete-block design assuming fixed treatment effects and random block effects. The treatments sums of squares were partitioned into a comparison of the control vs. all fertilizer treatments and the factorial combinations of the P rate and time of application treatments excluding the non-fertilized control treatment (because the time of application treatment did not apply to the control) that included effects of P rate (5), time of application (2), and their interaction. To better evaluate a possible interaction for the lower P application rates, we partitioned the sums of squares of the interaction to test if the difference between the mean of the 10- and 20-kg rates and the mean of higher rates was statistically similar for the fall and spring times of P application. When the main interaction or the described contrast for the two smallest P rates were not significant, the mean crop response to P across the two times of application was further studied by fitting various response models. Otherwise the models were fit to crop response for each time of application. We fit linear, quadratic, linear-plateau (LP), and quadraticplateau (QP) models usually used to describe crop response to fertilization using the procedures REG or NLIN of SAS. Curvilinear or LP models were chosen to represent crop response only when the fit resulted in significant smaller (P 0.10) residual error than the linear model as indicated by an F test of the residual sums of squares. In the few instances in which LP and QP models did not differ, we chose the model with the more reasonable fit based on observation of data points and model lines as suggested by others (Cerrato and

21 14 Blackmer, 1990). We also used the MIXED procedure of SAS similar to the one used by site to evaluate the average treatment effects on each crop measurement across the responsive sites. A site was classified as yield responsive when there was a grain yield response to P (at P 0.10) for one or both times of P application. RESULTS AND DISCUSSION Corn and Soybean Grain Yield Responses The main objective of this study was to evaluate the efficacy of fall and spring broadcast P fertilization for corn and soybean managed with no-tillage. For example, we wanted to determine if grain yield was affected by spring or fall P application. Such an assessment can be done only in sites where the crops responded to P application. Data in Table 3 show that grain yield responses were significant (P 0.10) to one or more P treatments at seven soybean sites (at P 0.05 at six of them) and four corn sites (at P 0.05 at all of them). Current Iowa State University STP interpretations for fields managed with any tillage system indicate (based on previous field research) that the probability of a yield response is 80, 65, 25, and 5% for the categories Very Low, Low, Optimum, and High, respectively (Sawyer et al., 2002). The lowest recommended amount of fertilizer is for the Optimum category, and the rate is designed to maintain STP. The three soil-test methods used usually agreed at classifying the sites into the interpretation categories, and when they did not the values were borderline between two categories and a different classification was

22 15 determined by only 2 mg P kg -1 or less. The values used to classify soils of the sites into soiltest interpretation categories shown in Table 2 are averages for a 15-cm depth, which is the criterion used in Iowa for no-tillage and all tillage systems (Sawyer et al., 2002, 2003). If we assume a probability 25% or larger of a large to small response in soils testing Very Low to Optimum, the three soil-test methods agreed at classifying the ten sites in these categories as yield responsive. The Bray-P1 and Mehlich-3 tests also agreed at classifying into the High class the responsive Site 13 (values were within 2 mg P kg -1 of the Optimum category) but the Olsen test classified it as Optimum. One possible explanation for the response to P fertilization at this site could be the high STP stratification found, because the top 7.5 cm of soil had nearly 200 % more P than the second 7.5 cm layer. The methods also agreed at classifying into the Very Low or Low categories two corn sites and two soybean sites where grain yield response to P was not observed (Sites 8, 12, 17, and 19). Soil-test P stratification (Table 2) did not explain a lack of response at these sites, and it might be explained by soil sampling error, within-site STP variability, or effects of undetermined site conditions that could have increased the soil P availability or limited the response. The time of broadcast P application did not affect (P 0.10) the corn or soybean grain yield response to P applied at any site (Table 3). The main interaction between time of P application and P rate was non-significant. Furthermore, no interaction was significant (P 0.10) when the sums of squares of the main interaction were partitioned to test differences in response for the two lowest P rates applied each season or the linear and quadratic general responsive trends for P applied each season (not shown). The only apparent statistical evidence for an interaction was observed at Site 4 for the 10- and 20-kg P rates at P = 0.08

23 16 because of a higher grain yield for the fall-applied 10-kg rate and a higher yield for the spring-applied 20-kg rate (Table 3). However, this is a biologically unreasonable result because nothing other than experimental error can explain an opposite season effect for the 10- and 20-kg rates. Therefore, we conclude that the effects of P application on grain yield were statistically similar for the two times of application at this site and all other responsive sites. The general lack of interaction effects determined the most practical way of presenting and discussing the yield responses to P. Although treatment interactions were not present, we show grain yield data for each time of P application and P rate for each site in Table 3. However, linear, quadratic, quadratic-plateau, and linear plateau models used to relate grain yield with P rate for each responsive site were fit to means across the two times of P application. Table 4 shows the model used and the P rate estimated to produce the maximum yield for each responsive site. At Sites 2, 4, 5, 7, and 20, the grain yield responses showed a linear response trend with a 23.8-, 24.1-, 9.9-, 3.18-, and 7.28-kg yield increment for every kg of P applied, respectively. The positive yield response to P fertilization at Sites 6 and 11 is attributed to the 10 kg P ha -1 rate based on the comparison of the control and the average of all fertilized plots because no model fit the small response. At Site 13 the yield response showed a linear-plateau trend and 20 kg P ha -1 maximized yield. At Sites 14, 15, and 16 the yield responses showed a quadratic-plateau response trend; and 17, 15, and 37 kg P ha -1 maximized yield, respectively. Mean grain yield responses across responsive sites and the two times of P application showed a quadratic-plateau response trend for both crops (Fig. 1). An analysis of variance of

24 17 P rate and time of P application effects indicate a non-significant interaction between P rate and the time of application, even for the low P rates. Therefore, the graphs for each crop show the estimated mean yields across the two times of application although the observed data points for each time of P application also are shown. Rates of 31 and 29 kg P ha -1 maximized yield of corn and soybean, respectively. The observed data for the corn graph seem to indicate a lower P efficiency for the 20-kg rate applied in the fall compared with the same rate applied in spring. However, an orthogonal comparison indicated a non-significant different (P 0.10) and, furthermore, responses for the 10- and 30-kg rates suggest that apparent difference is a random result. Similarly, very small apparent soybean yield differences in favor of the spring application of the 10- and 20-kg rates were not significant. Therefore, we conclude that the time of P application did not affect the corn and soybean yield response to P, and that on average across sites an almost similar P rate maximized the yield response for both crops. Corn and Soybean Early Dry Weight Responses There were early plant growth responses (P 0.10) to one or more P treatments at only two soybean sites and one corn site (Table 5), which were Sites 4, 8, and 19. The three sites tested Low or Very Low in STP (Bray-1 < 11 mg P kg -1, 0-15 cm sampling depth), although responses did not occur for all low-testing soils. It is noteworthy that only one site (Site 4, corn) responsive for early growth also showed a grain yield response. At Sites 4 and 8, both the time of P application main effect and the interaction between time of P application and P rate were not significant, which indicated similar response to P for both times of application.

25 18 At Site 4 a partition of the interaction sums of squares suggested that the difference in response to P for the low and high rates differed (at P 0.07) across seasons (Table 5). The DW data shows that the interaction probably is explained by a lower response to the springapplied 10-kg rate and greater response to the spring-applied 40-kg rate compared to similar fall-applied P rates. This result is difficult to explain and, therefore, we assume this was a random result. Linear, quadratic, and quadratic-plateau models were fit to early growth and P rate for each site to determine the P rate that maximize yield (Table 6). At Site 4, early plant growth showed a quadratic-plateau response trend and the rate that maximized plant early growth was 23 kg P ha -1. At Site 8, early growth showed a quadratic response, and 30 kg P ha -1 maximized yields. At Site 19 the interaction was significant and soybean early growth responses differed for the fall and spring times of P application. For fall, early growth showed a linear response with a slope of g DW per kg P, and for spring early growth showed a quadratic response and 26 kg P ha -1 maximized growth. Having a quadratic model showing a negative trend for early growth at the highest rates of spring-applied P, which explains the significant interaction, was not expected and cannot be explained satisfactorily. A partition of the interaction sums of squares for responses to the lowest P rates (10 and 20 kg P ha -1 ) indicated no significant difference between times of P application. There were apparent significant interaction effects between P rate and time of P application at Sites 14 and 17 that were biologically unreasonable and probably random results. At these sites there were no significant growth responses to P for data analyzed by season or for averages across seasons (P 0.10). At Site 14, a significant (P 0.07) partial interaction is the result of no obvious difference between times of application for the two low P rates but a difference

26 19 between seasons for the three high P rates with DW being greater for fall than for spring. At Site 17, a significant main interaction (P 0.09) is the result of higher DW for the fallapplied 30- and 40-kg P rates than for the 50-kg rate and an opposite trend for spring-applied P. Mean early growth responses across the three sites responsive to P across the two times of P application showed a quadratic-plateau response trend for corn and a linear-plateau response trend for soybean (Fig. 2). An analysis of variance of P rate and time of application effects also indicated a non-significant (P 0.10) interaction between P rate and the time of application, even for the low P rates (not shown). Rates of 24 and 21 kg P ha -1 maximized early growth of corn and soybean, respectively. The data points for the corn graph seem to indicate a lower P efficiency for the 10-kg rate applied in the spring compared with the same rate applied in fall, but an orthogonal comparison also indicated a non-significant difference and, furthermore, observation of responses for the control- and 20-kg rates suggest that apparent difference is a random result. For soybean, a small apparent early growth difference in favor of the spring application for the 10-, 20-, and 30-kg rates and in favor of the fall application for the 40- and 50-kg rates were not significant. Therefore, results by responsive site and across the responsive sites indicate no significant differences between fall and spring times of application and that, on average, an approximately similar P rate maximized early growth of corn and soybean.

27 20 Corn and Soybean Early P Concentration Responses There were plant P concentration responses (P 0.10) to P at four soybean sites and two corn sites (Table 7). All responsive sites tested Very Low or Low in STP (Bray-1 <16 mg P kg -1, 0-15 cm depth). None of these sites showed an early growth response and four were also responsive for grain yield (Sites 5, 14, 15, and 16). Only one site (Site 16) showed a significant (P 0.08) main interaction between time of application and P rate, which indicated a different response to P applied in fall or spring (Table 7). Furthermore, the interaction was also significant (P 0.02) when the sums of squares were partitioned to test differences in response to the two lowest P rates applied each season. There was an apparent significant (P 0.10) interaction between P rate and time of P application at Site 8, where there was no significant plant P concentration response to P for data analyzed by season or for averages across seasons. This statistical interaction is the result of high (apparently random) plant P concentration variation for P rates higher than 30 kg P ha -1 applied in fall or spring. Linear and quadratic-plateau models were fit to relate P concentration with P rate for each responsive site (to means of fall and spring times of application for all sites). This allowed us to identify the P rate that maximized P concentration in young plants. Table 8 shows the models used and the P rate estimated to produce the maximum P concentration. At Site 5, the early soybean P concentration response showed a linear trend. The positive response of early soybean P concentration to P fertilization at Site 14 is attributed to the 10 kg P ha -1 rate based on the orthogonal comparison of P concentration for the control and the average of all fertilized plots because no model fit the response. At Sites 15 and 16, the P

28 21 concentration responses of soybean (Site 15) and corn (Site 16) showed a linear trend. However, at Site 16 we needed to describe separately the corn P concentration responses for fall and spring because the interaction between time of application and P rate was significant. Fall and spring responses showed an increment of and mg P kg -1 for each unit of P applied, respectively, which indicates a higher response to P applied in spring. At Site 17, mean early corn P concentration showed a quadratic-plateau response trend, and 39 kg P ha -1 maximized P concentration. At Site 18, mean soybean P concentration showed a linear trend, with a slope of mg P kg -1 for each unit of P applied. This Site showed a smaller increment in the mg of P kg -1 for each unit of P applied than Site 15. There was a significant (P 0.10) time of P application main effect at Site 2 (data not shown) that was difficult to explain because there was no significant response to P for data by season or for averages across seasons (P 0.10) and the interaction between season of P application and P rate was not significant when tested for all P rates or the lowest P rates. The mean P concentration response across the two times of P application and across the two corn responsive sites and the four soybean responsive sites was linear for both crops (Fig. 3). An analysis of variance for these averages for each crop indicated a non-significant interaction between P rate and the time of application, even for the lowest P rates. Data for soybean seem to indicate a greater P efficiency for the 10- and 20-kg rates applied in spring and a greater efficiency for the 40-kg and 50-kg rates applied in the fall, but both the interaction and orthogonal comparisons indicated non-significant differences (P 0.10). Similarly for corn, a non-significant interaction allows us to dismiss apparent and unreasonable greater efficiency for the 20- and 30-kg rates applied in fall and for the 10- and

29 22 40-kg rates applied in spring. Therefore, the averages across responsive sites indicate no significant effect for the time of P application for both crops and linear early plant P concentration responses to P by both crops. Data in Fig. 3 do suggest, however, that in the corn early P concentration response per kg of applied P was higher than for soybean. Corn and Soybean Early P Uptake Responses Early plant P uptake responses (P 0.10) to P were observed at four soybean sites (at 0.05 at three of them) and three corn sites (at 0.05 in all of them) (Table 9). Three of the P uptake responsive sites also were responsive in early growth (sites 4, 8, and 19), and three also were responsive in grain yield (Sites 4, 5, and 16). Four of the P uptake responsive sites also were responsive to P concentration (Sites 5, 16, 17, and 18). All this sites tested Very Low to a value borderline between Low and Optimum for STP (Bray-1 <17 mg P kg -1, 0-15 cm depth). The P uptake responses differed for P applied in fall or spring (P 0.10) at Site 19 as indicated by a significant interaction between time of P application and P rate (Table 9). There was an apparent significant (P 0.10) interaction between P rate and time of P application at Site 13, where there was no significant uptake response for data analyzed by season or for averages across seasons. This statistical interaction is the result of high random plant P uptake variation that does not have a reasonable biological explanation Linear, quadratic, and quadratic-plateau models were fit to mean early P uptake response to P across fall and spring times of application for each responsive site where there was no interaction between time of application and P rate (all responsive sites) in order to identify the P rate that maximized P uptake (Table 10). The positive responses to P

30 23 fertilization at Sites 4, 5, and 17 were attributed to the 10 kg P ha -1 rate based on the orthogonal comparison of P uptake for the control and the average of all fertilized plots because no model fit the responses. At Site 8, the P uptake responses showed a quadratic response trend, and 28 kg P ha -1 maximized P uptake. At Site 16, the response was linear with a 0.13 mg plant -1 increment for every kg ha -1 of P applied. At Site 18, the P uptake response showed a linear response. At Site 19 the interaction between time of P application and P rate was significant, so the responses were described separately for fall- and spring-applied P. The P uptake response for this Site showed a linear response for fall-applied P and a quadratic response for springapplied P with an estimated maximum for the 27 kg P ha -1 rate. The interaction is easily explained by the data in Table 9 and calculations from the equations in Table 10, but we do not understand the agronomic reasons for such a differential response. A similar uptake maximum (7.8 mg P plant -1 ) was achieved with a much higher fall-applied P rate (50 kg P ha - 1, the highest rate applied) than with the spring-applied P (27 kg P ha -1 ), which would suggest a higher efficiency of spring-applied P. However, higher spring-applied P reduced P uptake. This response tended to follow early growth responses, and neither has a reasonable explanation. There was a significant (P 0.10) difference between the P applied in the fall and in spring in the P uptake response at Site 12. Data (Table 9) seems to indicate a lower P efficiency for the mean of the spring rates, and a higher P efficiency for the 20-kg rate applied in the fall compared with the same rate applied in the spring, but the season by P rate interaction, and the interaction season by the two lower rates, indicated a non-significant difference (P 0.10).

31 24 Figure 4 shows the mean early P uptake responses across the two times of P application and across the three corn responsive sites and the four soybean responsive sites. An analysis of variance for these averages for each crop indicated a non-significant interaction between P rate and the time of application, even for the lowest P rates. The mean corn P uptake response was linear, but the mean soybean response followed a linear-plateau trend and 14 kg P ha -1 maximized soybean early P uptake. Therefore, the averages across responsive sites indicated no significant effect of the time of P application on early P uptake for both crops but showed a clear difference between crops in early P uptake response, being greater and to a higher P rate for corn. Summary Discussion Significant crop responses to P were observed at 11 sites for grain yield, three sites for early growth, six sites for early P concentration, and seven sites for early P uptake. The time of broadcast P application did not affect grain yield at any site. Data showed that both corn and soybean showed a grain yield response to P fertilization in soils testing Optimum or less in P, except for one soybean site that tested Optimum according to the Olsen method and borderline between the Optimum and High classes according to the Bray-P1 and Mehlich-3 methods. On average for the sites that were responsive to P fertilization, the P rates that maximized grain yield were 31 and 29 kg P ha -1 for corn and soybean, respectively. Several sites that showed a grain yield response to P did not show a response of early plant growth, P concentration, or P uptake. The P rates that on average maximized early corn and soybean growth response to P fertilization differed only by 3 kg P ha -1. However, although the

32 25 response of early plant P concentration was linear for both crops, corn showed a greater response to each kg of P applied than soybean (5.97 and 3.46 g P kg -1 per kg of applied P, respectively). The greater difference between crops was seen in the P uptake, however. For soybean, the maximum response was obtained with 14 kg P ha -1 applied, whereas corn showed a linear response. The time of broadcast P application did not affect the corn or soybean grain yield responses at any site. There were small and inconsistent or unreasonable time of P application effects at one site for early growth, one site for early P concentration, and one site for early P uptake. We had theorized that one possible reason for a lack of differences between P placement methods shown in previous Iowa studies could have been explained by broadcast P application usually made in the fall, 4 to 5 months before planting the crops. Application of P in the fall could provide sufficient time for P fertilizer granules or dissolving P to move below the residue cover mainly as a result of freezing or thawing, snow, and rain. An extended period of time between P application and seeding growth would not pose a problem because Iowa soils tend to retain applied P but their chemical and mineralogical properties do not result in the significant change to plant-unavailable forms observed in soils of other regions and, moreover, research has demonstrated that no-till management reduces P sorption in the shallowest soil layers (Guertal et al., 1991). On the other hand, P fertilizer applied in the spring at or shortly before planting might not have enough time to move under the residue and into the shallowest soil layers so it can not be absorbed by plant roots early in the season. The results of the study did not confirm our theory, and demonstrated no difference between fall and spring broadcast P application. This result agrees with starter

33 26 fertilization research with corn in Iowa (Kaiser et al., 2005), which showed no yield response to starter P when broadcast P had been applied although early plant growth often was increased. CONCLUSIONS Phosphorus fertilization increased grain yield at four corn sites and seven soybean sites, early plant growth at one corn site and two soybean sites, early plant P concentration at two corn sites and four soybean sites, and early P uptake at three corn sites and four soybean sites. The grain yield responses were observed at sites testing Optimum or less in STP measured at a 15-cm depth, except for one site classified into the High category by the Bray- 1 and Mehlich-3 soil-test methods (within 2 mg P kg -1 of the Optimum category) but Optimum by the Olsen method. Phosphorus did not increase grain yield at three sites where STP was classified Very Low or Low by three soil-test methods according to results for samples taken from a 7.5- or 15-cm depth. Sites showing a grain yield response seldom showed early plant growth, P concentration, or P uptake responses. Study of crop responses at the responsive sites showed no large or consistent differences in corn or soybean responses to P fertilizer broadcast in the fall or spring. The time of P application did not affect grain yield or early plant growth. The time of P application affected early plant P concentration only at one corn site, where the fall application was more efficient than the spring application for two lowest P rates but was less